import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
sns.set_style('whitegrid')
plt.style.use('bmh')

import warnings
warnings.filterwarnings('ignore')

from IPython.core.interactiveshell import InteractiveShell
InteractiveShell.ast_node_interactivity = 'all'
%config InlineBackend.figure_format='retina'

bank_df = pd.read_excel(r'C:\Data\Banking Data.xlsx')

bank_df

	banking marketing	Unnamed: 1	Unnamed: 2	Unnamed: 3	Unnamed: 4	Unnamed: 5	Unnamed: 6	Unnamed: 7	Unnamed: 8	Unnamed: 9	Unnamed: 10	Unnamed: 11	Unnamed: 12	Unnamed: 13	Unnamed: 14	Unnamed: 15	Unnamed: 16	Unnamed: 17	Unnamed: 18
0	customer id and age.	NaN	Customer salary and balance.	NaN	Customer marital status and job with education...	NaN	particular customer before targeted or not	NaN	Loan types: loans or housing loans	NaN	Contact type	NaN	month of contact	duration of call	NaN	NaN	NaN	outcome of previous contact	response of customer after call happned
1	customerid	age	salary	balance	marital	jobedu	targeted	default	housing	loan	contact	day	month	duration	campaign	pdays	previous	poutcome	response
2	1	58	100000	2143	married	management,tertiary	yes	no	yes	no	unknown	5	may, 2017	261 sec	1	-1	0	unknown	no
3	2	44	60000	29	single	technician,secondary	yes	no	yes	no	unknown	5	may, 2017	151 sec	1	-1	0	unknown	no
4	3	33	120000	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	5	may, 2017	76 sec	1	-1	0	unknown	no
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
45208	45207	51	60000	825	married	technician,tertiary	yes	no	no	no	cellular	17	nov, 2017	16.2833333333333 min	3	-1	0	unknown	yes
45209	45208	71	55000	1729	divorced	retired,primary	yes	no	no	no	cellular	17	nov, 2017	7.6 min	2	-1	0	unknown	yes
45210	45209	72	55000	5715	married	retired,secondary	yes	no	no	no	cellular	17	nov, 2017	18.7833333333333 min	5	184	3	success	yes
45211	45210	57	20000	668	married	blue-collar,secondary	yes	no	no	no	telephone	17	nov, 2017	8.46666666666667 min	4	-1	0	unknown	no
45212	45211	37	120000	2971	married	entrepreneur,secondary	yes	no	no	no	cellular	17	nov, 2017	6.01666666666667 min	2	188	11	other	no

45213 rows × 19 columns

Audjest Column name

bank_df.columns

Index(['banking marketing', 'Unnamed: 1', 'Unnamed: 2', 'Unnamed: 3',
       'Unnamed: 4', 'Unnamed: 5', 'Unnamed: 6', 'Unnamed: 7', 'Unnamed: 8',
       'Unnamed: 9', 'Unnamed: 10', 'Unnamed: 11', 'Unnamed: 12',
       'Unnamed: 13', 'Unnamed: 14', 'Unnamed: 15', 'Unnamed: 16',
       'Unnamed: 17', 'Unnamed: 18'],
      dtype='object')

list(bank_df[1:2].values)

[array(['customerid', 'age', 'salary', 'balance', 'marital', 'jobedu',
        'targeted', 'default', 'housing', 'loan', 'contact', 'day',
        'month', 'duration', 'campaign', 'pdays', 'previous', 'poutcome',
        'response'], dtype=object)]

bank_df.columns = ('customer_id', 'customer_age', 'salary', 'balance', 'marital', 'job_edu',
        'targeted_before_or_not', 'default', 'housing_loan', 'loan', 'contact', 'contact_day',
        'contact_month', 'durationof_call', 'campaign', 'pdays', 'previous', 'prv_outcome',
        'response_after_call')

bank_df.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	contact_day	contact_month	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call
0	customer id and age.	NaN	Customer salary and balance.	NaN	Customer marital status and job with education...	NaN	particular customer before targeted or not	NaN	Loan types: loans or housing loans	NaN	Contact type	NaN	month of contact	duration of call	NaN	NaN	NaN	outcome of previous contact	response of customer after call happned
1	customerid	age	salary	balance	marital	jobedu	targeted	default	housing	loan	contact	day	month	duration	campaign	pdays	previous	poutcome	response
2	1	58	100000	2143	married	management,tertiary	yes	no	yes	no	unknown	5	may, 2017	261 sec	1	-1	0	unknown	no
3	2	44	60000	29	single	technician,secondary	yes	no	yes	no	unknown	5	may, 2017	151 sec	1	-1	0	unknown	no
4	3	33	120000	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	5	may, 2017	76 sec	1	-1	0	unknown	no

bank_df=bank_df.drop([0,1])

bank_df.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	contact_day	contact_month	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call
2	1	58	100000	2143	married	management,tertiary	yes	no	yes	no	unknown	5	may, 2017	261 sec	1	-1	0	unknown	no
3	2	44	60000	29	single	technician,secondary	yes	no	yes	no	unknown	5	may, 2017	151 sec	1	-1	0	unknown	no
4	3	33	120000	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	5	may, 2017	76 sec	1	-1	0	unknown	no
5	4	47	20000	1506	married	blue-collar,unknown	no	no	yes	no	unknown	5	may, 2017	92 sec	1	-1	0	unknown	no
6	5	33	0	1	single	unknown,unknown	no	no	no	no	unknown	5	may, 2017	198 sec	1	-1	0	unknown	no

bank_df.reset_index(drop=True,inplace=True)

bank_df.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	contact_day	contact_month	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call
0	1	58	100000	2143	married	management,tertiary	yes	no	yes	no	unknown	5	may, 2017	261 sec	1	-1	0	unknown	no
1	2	44	60000	29	single	technician,secondary	yes	no	yes	no	unknown	5	may, 2017	151 sec	1	-1	0	unknown	no
2	3	33	120000	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	5	may, 2017	76 sec	1	-1	0	unknown	no
3	4	47	20000	1506	married	blue-collar,unknown	no	no	yes	no	unknown	5	may, 2017	92 sec	1	-1	0	unknown	no
4	5	33	0	1	single	unknown,unknown	no	no	no	no	unknown	5	may, 2017	198 sec	1	-1	0	unknown	no

Data cleaning

bank_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 45211 entries, 0 to 45210
Data columns (total 19 columns):
 #   Column                  Non-Null Count  Dtype 
---  ------                  --------------  ----- 
 0   customer_id             45211 non-null  object
 1   customer_age            45191 non-null  object
 2   salary                  45211 non-null  object
 3   balance                 45211 non-null  object
 4   marital                 45211 non-null  object
 5   job_edu                 45211 non-null  object
 6   targeted_before_or_not  45211 non-null  object
 7   default                 45211 non-null  object
 8   housing_loan            45211 non-null  object
 9   loan                    45211 non-null  object
 10  contact                 45211 non-null  object
 11  contact_day             45211 non-null  object
 12  contact_month           45161 non-null  object
 13  durationof_call         45211 non-null  object
 14  campaign                45211 non-null  object
 15  pdays                   45211 non-null  object
 16  previous                45211 non-null  object
 17  prv_outcome             45211 non-null  object
 18  response_after_call     45181 non-null  object
dtypes: object(19)
memory usage: 6.6+ MB

bank_df.customer_id=bank_df.customer_id.astype(int)
bank_df.customer_age=bank_df.customer_age.astype(float)
bank_df.balance = bank_df.balance.astype(int)
bank_df.contact_day = bank_df.contact_day.astype(int)
bank_df.campaign = bank_df.campaign.astype(int)
bank_df.pdays = bank_df.pdays.astype(int)
bank_df.previous = bank_df.previous.astype(int)

bank_df.salary=bank_df['salary'].replace('?',np.nan)

bank_df.salary = bank_df.salary.astype(float)

bank_df.isna().sum()

customer_id                0
customer_age              20
salary                    62
balance                    0
marital                    0
job_edu                    0
targeted_before_or_not     0
default                    0
housing_loan               0
loan                       0
contact                    0
contact_day                0
contact_month             50
durationof_call            0
campaign                   0
pdays                      0
previous                   0
prv_outcome                0
response_after_call       30
dtype: int64

# Dealing with salary cloumn to fill null values
bank_df.salary.mean()
bank_df.salary.median()
bank_df.salary.plot(kind='box')
# Because of no outliers we can fill null values with mean
bank_df.salary.replace(np.nan,bank_df.salary.mean(),inplace=True)

57084.45369775632

60000.0

<AxesSubplot:>

bank_df.isna().sum() 

customer_id                0
customer_age              20
salary                     0
balance                    0
marital                    0
job_edu                    0
targeted_before_or_not     0
default                    0
housing_loan               0
loan                       0
contact                    0
contact_day                0
contact_month             50
durationof_call            0
campaign                   0
pdays                      0
previous                   0
prv_outcome                0
response_after_call       30
dtype: int64

Deal with nan value

#1) Dealing with  null values of customer column age 
bank_df.customer_age.mean()
bank_df.customer_age.median()

bank_df.customer_age=bank_df.customer_age.fillna(bank_df.customer_age.mean())
# replacing null values with mean of customer age 

40.93565090394105

39.0

bank_df.customer_age.isna().sum()

0

#2) dealing with customer contact month column
# Filling the null values with interpolate
bank_df.contact_month.interpolate(method='ffill',limit_direction='forward',inplace=True)
bank_df.contact_month.isna().sum()

0

bank_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 45211 entries, 0 to 45210
Data columns (total 19 columns):
 #   Column                  Non-Null Count  Dtype  
---  ------                  --------------  -----  
 0   customer_id             45211 non-null  int32  
 1   customer_age            45211 non-null  float64
 2   salary                  45211 non-null  float64
 3   balance                 45211 non-null  int32  
 4   marital                 45211 non-null  object 
 5   job_edu                 45211 non-null  object 
 6   targeted_before_or_not  45211 non-null  object 
 7   default                 45211 non-null  object 
 8   housing_loan            45211 non-null  object 
 9   loan                    45211 non-null  object 
 10  contact                 45211 non-null  object 
 11  contact_day             45211 non-null  int32  
 12  contact_month           45211 non-null  object 
 13  durationof_call         45211 non-null  object 
 14  campaign                45211 non-null  int32  
 15  pdays                   45211 non-null  int32  
 16  previous                45211 non-null  int32  
 17  prv_outcome             45211 non-null  object 
 18  response_after_call     45181 non-null  object 
dtypes: float64(2), int32(6), object(11)
memory usage: 5.5+ MB

#3) dealing with null values of (response after call) column 
bank_df.response_after_call.astype(str).mode()

0    no
Name: response_after_call, dtype: object

bank_df.response_after_call.interpolate(method='ffill',limit_direction='forward',inplace=True)

# Merging day column with month column of dataframe
bank_df['contact_Date'] = bank_df['contact_day'].astype(str)+bank_df['contact_month']

bank_df.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	contact_day	contact_month	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call	contact_Date
0	1	58.0	100000.0	2143	married	management,tertiary	yes	no	yes	no	unknown	5	may, 2017	261 sec	1	-1	0	unknown	no	5may, 2017
1	2	44.0	60000.0	29	single	technician,secondary	yes	no	yes	no	unknown	5	may, 2017	151 sec	1	-1	0	unknown	no	5may, 2017
2	3	33.0	120000.0	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	5	may, 2017	76 sec	1	-1	0	unknown	no	5may, 2017
3	4	47.0	20000.0	1506	married	blue-collar,unknown	no	no	yes	no	unknown	5	may, 2017	92 sec	1	-1	0	unknown	no	5may, 2017
4	5	33.0	0.0	1	single	unknown,unknown	no	no	no	no	unknown	5	may, 2017	198 sec	1	-1	0	unknown	no	5may, 2017

# Changing contact_date cloumn to dateformat
from datetime import datetime
bank_df['Con_Date'] = pd.to_datetime(bank_df.contact_Date,format='%d%b, %Y')

# Droping unrequired cloumn of date
bank_df.drop(['contact_day','contact_month','contact_Date'],axis=1,inplace=True)

bank_df.to_csv(r'C:\Data\Banking_cleaned_Data.csv')

bank_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 45211 entries, 0 to 45210
Data columns (total 18 columns):
 #   Column                  Non-Null Count  Dtype         
---  ------                  --------------  -----         
 0   customer_id             45211 non-null  int32         
 1   customer_age            45211 non-null  float64       
 2   salary                  45211 non-null  float64       
 3   balance                 45211 non-null  int32         
 4   marital                 45211 non-null  object        
 5   job_edu                 45211 non-null  object        
 6   targeted_before_or_not  45211 non-null  object        
 7   default                 45211 non-null  object        
 8   housing_loan            45211 non-null  object        
 9   loan                    45211 non-null  object        
 10  contact                 45211 non-null  object        
 11  durationof_call         45211 non-null  object        
 12  campaign                45211 non-null  int32         
 13  pdays                   45211 non-null  int32         
 14  previous                45211 non-null  int32         
 15  prv_outcome             45211 non-null  object        
 16  response_after_call     45211 non-null  object        
 17  Con_Date                45211 non-null  datetime64[ns]
dtypes: datetime64[ns](1), float64(2), int32(5), object(10)
memory usage: 5.3+ MB

bank_df.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call	Con_Date
0	1	58.0	100000.0	2143	married	management,tertiary	yes	no	yes	no	unknown	261 sec	1	-1	0	unknown	no	2017-05-05
1	2	44.0	60000.0	29	single	technician,secondary	yes	no	yes	no	unknown	151 sec	1	-1	0	unknown	no	2017-05-05
2	3	33.0	120000.0	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	76 sec	1	-1	0	unknown	no	2017-05-05
3	4	47.0	20000.0	1506	married	blue-collar,unknown	no	no	yes	no	unknown	92 sec	1	-1	0	unknown	no	2017-05-05
4	5	33.0	0.0	1	single	unknown,unknown	no	no	no	no	unknown	198 sec	1	-1	0	unknown	no	2017-05-05

bank_df.describe()

	customer_id	customer_age	salary	balance	campaign	pdays	previous
count	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000
mean	22606.000000	40.935651	57084.453698	1362.272058	2.763841	40.197828	0.580323
std	13051.435847	10.616848	32016.099611	3044.765829	3.098021	100.128746	2.303441
min	1.000000	18.000000	0.000000	-8019.000000	1.000000	-1.000000	0.000000
25%	11303.500000	33.000000	20000.000000	72.000000	1.000000	-1.000000	0.000000
50%	22606.000000	39.000000	60000.000000	448.000000	2.000000	-1.000000	0.000000
75%	33908.500000	48.000000	70000.000000	1428.000000	3.000000	-1.000000	0.000000
max	45211.000000	95.000000	120000.000000	102127.000000	63.000000	871.000000	275.000000

Data visualization

Univarieate Analysis

1)Histogram

plt.figure(figsize=(12,5))
plt.hist(bank_df.customer_age,edgecolor = 'b', color= 'orange');

Insight

1)We can see that maximum count of customer of age who have account in bank is bitween 25 to 40.

2)After 55 age of customers,there is very minimum no of account.

plt.figure(figsize=(12,5))
plt.hist(bank_df.salary,edgecolor = 'b', color= 'orange');

Observation= Maximum no of salary of customers are 10k-20k, 50k-70k and 1lakh-1.20lakh

2) Box plot

plt.figure(figsize=(12,5))
sns.boxplot(data=bank_df,x='salary');

observation = Range of salary is 20k to 70k and there is no outliers in salary and median of salary is 60k.

plt.figure(figsize=(12,5))
sns.boxplot(data=bank_df,x='balance');

observation = Maximum customers save there salary bitween 0 to 1500 rs and there are some customers who save large amount of there salary.

plt.figure(figsize=(12,5))
sns.boxplot(data=bank_df,x='customer_age');

observation= Age of customers bitween 32 to 47 have maximum no of account in this bank

plt.figure(figsize=(12,5))
sns.boxplot(data=bank_df,x='campaign');

#starting point
bank_df1 = pd.read_csv(r'C:\Data\Banking_cleaned_Data.csv')

from datetime import datetime

bank_df1['Con_Date'] = pd.to_datetime(bank_df1.Con_Date)
bank_df1.drop('Unnamed: 0',axis=1,inplace=True)

bank_df1.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 45211 entries, 0 to 45210
Data columns (total 18 columns):
 #   Column                  Non-Null Count  Dtype         
---  ------                  --------------  -----         
 0   customer_id             45211 non-null  int64         
 1   customer_age            45211 non-null  float64       
 2   salary                  45211 non-null  float64       
 3   balance                 45211 non-null  int64         
 4   marital                 45211 non-null  object        
 5   job_edu                 45211 non-null  object        
 6   targeted_before_or_not  45211 non-null  object        
 7   default                 45211 non-null  object        
 8   housing_loan            45211 non-null  object        
 9   loan                    45211 non-null  object        
 10  contact                 45211 non-null  object        
 11  durationof_call         45211 non-null  object        
 12  campaign                45211 non-null  int64         
 13  pdays                   45211 non-null  int64         
 14  previous                45211 non-null  int64         
 15  prv_outcome             45211 non-null  object        
 16  response_after_call     45211 non-null  object        
 17  Con_Date                45211 non-null  datetime64[ns]
dtypes: datetime64[ns](1), float64(2), int64(5), object(10)
memory usage: 6.2+ MB

bank_df1.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call	Con_Date
0	1	58.0	100000.0	2143	married	management,tertiary	yes	no	yes	no	unknown	261 sec	1	-1	0	unknown	no	2017-05-05
1	2	44.0	60000.0	29	single	technician,secondary	yes	no	yes	no	unknown	151 sec	1	-1	0	unknown	no	2017-05-05
2	3	33.0	120000.0	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	76 sec	1	-1	0	unknown	no	2017-05-05
3	4	47.0	20000.0	1506	married	blue-collar,unknown	no	no	yes	no	unknown	92 sec	1	-1	0	unknown	no	2017-05-05
4	5	33.0	0.0	1	single	unknown,unknown	no	no	no	no	unknown	198 sec	1	-1	0	unknown	no	2017-05-05

3)Count plot

plt.figure(figsize=(12,5))
sns.countplot(data=bank_df1,x='loan');

Maximum customer do use to take loan from bank apprx 7000 taken loan from bank out of 35000

plt.figure(figsize=(12,5))
sns.countplot(data=bank_df1,x='housing_loan');

Maximum customer do use to take loan from bank apprx 7000 taken loan from bank out of 35000

plt.figure(figsize=(12,5))
sns.countplot(data=bank_df1,x='housing_loan');

obseravation =1) 25000 customer take house loan out of 45211 customers 2) 20000 cutomer do not take house loan

plt.figure(figsize=(12,5))
sns.countplot(data=bank_df1,x='marital');

observation = 1)28k customers are maried

2)13k customers are single

3)5k customers are divorced

plt.figure(figsize=(12,5))
sns.countplot(data=bank_df1,x='response_after_call');

only 8% customer responed after call out of 100%

4) scatter plot

sns.scatterplot(data =bank_df1, x = 'balance', y= 'customer_age');

In scatter plot we can see that maximum peopal save there money is below 20000 and the age of customer between 30 to 60 saves more money.

sns.scatterplot(data = bank_df1, x = 'salary', y= 'customer_age');

5)pair plot

sns.pairplot(bank_df1);

6) Hexin plot

bank_df1.plot.hexbin(x='salary',
                    y='customer_age',
                    reduce_C_function=np.sum,
                    gridsize=10)

<AxesSubplot:xlabel='salary', ylabel='customer_age'>

7) Bar plot

sns.barplot(data = bank_df1, x = 'marital', y = 'salary', ci = None, palette='bright');

maximum no. of divorced peopal have salary above 60,000

sns.barplot(data = bank_df1, x = 'marital', y = 'balance', ci = None, palette='bright');

Maximum no of maried peopal save there salary.

sns.barplot(data = bank_df1, x = 'housing_loan', y = 'salary', ci = None, palette='bright');

8) Distribution plot

plt.figure(dpi = 200)
sns.displot(data = bank_df1, x = 'salary', height = 5, aspect = 15/5,kde = True);

<Figure size 1200x800 with 0 Axes>

maximum no of customer's earning salary are 20k, 60k and 1lakh.

plt.figure(dpi = 200)
sns.displot(data = bank_df1, x = 'balance', height = 5, aspect = 15/5,kde = True);

<Figure size 1200x800 with 0 Axes>

approx 2000 customers save there salary up to 1500

9) Heat map

sns.heatmap(bank_df1.corr(), cmap = 'rainbow', annot=True);

Statistical manipulation

1. Find the correlation between the columns and draw the observations from it.

bank_df1.corr()

	customer_id	customer_age	salary	balance	campaign	pdays	previous
customer_id	1.000000	0.014902	0.040000	0.073639	-0.102884	0.437729	0.271098
customer_age	0.014902	1.000000	0.026233	0.097744	0.004851	-0.023765	0.001329
salary	0.040000	0.026233	1.000000	0.056322	0.015300	-0.015560	0.014186
balance	0.073639	0.097744	0.056322	1.000000	-0.014578	0.003435	0.016674
campaign	-0.102884	0.004851	0.015300	-0.014578	1.000000	-0.088628	-0.032855
pdays	0.437729	-0.023765	-0.015560	0.003435	-0.088628	1.000000	0.454820
previous	0.271098	0.001329	0.014186	0.016674	-0.032855	0.454820	1.000000

Observation= 1)Costomer id is directly proportional to previous day 2)Maximum column are not proportonal to each other, show almost 0 relation with each other

2. What is the mean age and duration time of the customers with respect to every column?

bank_df1.describe()

	customer_id	customer_age	salary	balance	campaign	pdays	previous
count	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000	45211.000000
mean	22606.000000	40.935651	57084.453698	1362.272058	2.763841	40.197828	0.580323
std	13051.435847	10.616848	32016.099611	3044.765829	3.098021	100.128746	2.303441
min	1.000000	18.000000	0.000000	-8019.000000	1.000000	-1.000000	0.000000
25%	11303.500000	33.000000	20000.000000	72.000000	1.000000	-1.000000	0.000000
50%	22606.000000	39.000000	60000.000000	448.000000	2.000000	-1.000000	0.000000
75%	33908.500000	48.000000	70000.000000	1428.000000	3.000000	-1.000000	0.000000
max	45211.000000	95.000000	120000.000000	102127.000000	63.000000	871.000000	275.000000

3. Find the mean and median of every column response wise and draw the observations.

bank_df1.mean()
bank_df1.median()

customer_id     22606.000000
customer_age       40.935651
salary          57084.453698
balance          1362.272058
campaign            2.763841
pdays              40.197828
previous            0.580323
dtype: float64

customer_id     22606.0
customer_age       39.0
salary          60000.0
balance           448.0
campaign            2.0
pdays              -1.0
previous            0.0
dtype: float64

Observation= In customer's balance and previous day column there is a large difference b/n mean and meadian,it means this column consist maximum no. of outliears

4)Show that the columns are following the Normal Distribution or not, if not following try to convert it non-normal to normal.

# salary 
sns.displot(data=bank_df1,x='salary',kde=True)

<seaborn.axisgrid.FacetGrid at 0x1c9d2e12c70>

np.mean(bank_df1.salary)
np.std(bank_df1.salary)

np.mean(bank_df1.salary)+np.std(bank_df1.salary)

np.mean(bank_df1.salary)-np.std(bank_df1.salary)

57084.453697756246

32015.7455352698

89100.19923302604

25068.708162486444

Converting salary column to normal disritution

bank_df1.salary.skew()

0.13913623752811108

np.log(bank_df1.salary).skew()

nan

sns.displot(np.log(bank_df1.salary), kde = True);

2 Age of customer

# Age
sns.displot(data=bank_df1,x='customer_age',kde=True)

<seaborn.axisgrid.FacetGrid at 0x1c9ce15ffd0>

bank_df1.customer_age.skew()

0.6851529717199948

np.log(bank_df1.customer_age).skew()

0.09882818718129278

sns.displot(np.log(bank_df1.customer_age), kde = True);

Balance column

sns.displot(data=bank_df1,x='balance',kde=True)

<seaborn.axisgrid.FacetGrid at 0x1c9d488abb0>

bank_df1.balance.skew()

8.360308326166326

converting to normal distribution

np.log(bank_df1.balance).skew()

nan

sns.displot(np.log(bank_df1.balance),kde=True)

<seaborn.axisgrid.FacetGrid at 0x1c9d2c0e970>

Now its looking normal distribution bell like curve

5)Find the Best features using correlation and Chi-square test.

Chi-square test

Trying to find relation between Job and house loan

# Fixing job column
bank_df1['Job'] = bank_df1.job_edu.apply(lambda x: x.split(',')[0])
bank_df1['Education'] = bank_df1.job_edu.apply(lambda x: x.split(',')[1])

bank_df1.head()

	customer_id	customer_age	salary	balance	marital	job_edu	targeted_before_or_not	default	housing_loan	loan	contact	durationof_call	campaign	pdays	previous	prv_outcome	response_after_call	Con_Date	Job	Education
0	1	58.0	100000.0	2143	married	management,tertiary	yes	no	yes	no	unknown	261 sec	1	-1	0	unknown	no	2017-05-05	management	tertiary
1	2	44.0	60000.0	29	single	technician,secondary	yes	no	yes	no	unknown	151 sec	1	-1	0	unknown	no	2017-05-05	technician	secondary
2	3	33.0	120000.0	2	married	entrepreneur,secondary	yes	no	yes	yes	unknown	76 sec	1	-1	0	unknown	no	2017-05-05	entrepreneur	secondary
3	4	47.0	20000.0	1506	married	blue-collar,unknown	no	no	yes	no	unknown	92 sec	1	-1	0	unknown	no	2017-05-05	blue-collar	unknown
4	5	33.0	0.0	1	single	unknown,unknown	no	no	no	no	unknown	198 sec	1	-1	0	unknown	no	2017-05-05	unknown	unknown

import scipy.stats as stats

data_crosstab = pd.crosstab(bank_df1.Job,bank_df1.housing_loan,margins=True,margins_name='Total')

data_crosstab

housing_loan	no	yes	Total
Job
admin.	1989	3182	5171
blue-collar	2684	7048	9732
entrepreneur	618	869	1487
housemaid	842	398	1240
management	4780	4678	9458
retired	1773	491	2264
self-employed	814	765	1579
services	1388	2766	4154
student	689	249	938
technician	3482	4115	7597
unemployed	760	543	1303
unknown	262	26	288
Total	20081	25130	45211

# significance level
alpha = 0.05

rows = bank_df1.Job.unique()
columns = bank_df1.housing_loan.unique()

print(rows, columns)

['management' 'technician' 'entrepreneur' 'blue-collar' 'unknown'
 'retired' 'admin.' 'services' 'self-employed' 'unemployed' 'housemaid'
 'student'] ['yes' 'no']

# Finding chi square value
chi_square = []
for i in columns:
    for j in rows:
        O = data_crosstab[i][j]
        E = data_crosstab[i]['Total'] * data_crosstab['Total'][j] / data_crosstab['Total']['Total']
        chi_square.append((O-E)**2/E)

chi_square

[63.794863997804036,
 2.7470234319781315,
 2.182112082553112,
 496.3480499077622,
 112.30424784699052,
 467.99235587860153,
 32.95365538421202,
 90.47094588179588,
 14.463534136876882,
 45.36255439859088,
 123.06371715292835,
 142.29418193919253,
 79.8349152066538,
 3.4377122078387754,
 2.7307642365698777,
 621.1456846861244,
 140.54109598102045,
 585.6604702569223,
 41.23924903168408,
 113.21820975098515,
 18.100125136184253,
 56.768138640336076,
 154.00583696295448,
 178.07145023315113]

chi_square = np.sum(chi_square)

print(chi_square)

3588.7308943697108

# The critical value approach
print("\n--------------------------------------------------------------------------------------")
print(" The critical value approach to hypothesis testing in the decision rule")
critical_value = stats.chi2.ppf(1-alpha, (len(rows)-1)*(len(columns)-1))

if chi_square > critical_value:
    conclusion = "Null Hypothesis is rejected."
else:
    conclusion = "Failed to reject the null hypothesis."
        
print("chisquare-score is:", chi_square, " and critical value is:", critical_value)
print(conclusion)

--------------------------------------------------------------------------------------
 The critical value approach to hypothesis testing in the decision rule
chisquare-score is: 3588.7308943697108  and critical value is: 19.67513757268249
Null Hypothesis is rejected.

6)Find the probabilities with respect to the job role and education with customer responses.

# Making contingency table JOb with respect to response after call
contigency_table = pd.crosstab(bank_df1.Job,bank_df1.response_after_call,margins=True,margins_name='Total')

contigency_table

response_after_call	no	yes	Total
Job
admin.	4540	631	5171
blue-collar	9024	708	9732
entrepreneur	1363	124	1487
housemaid	1131	109	1240
management	8157	1301	9458
retired	1747	517	2264
self-employed	1392	187	1579
services	3784	370	4154
student	669	269	938
technician	6757	840	7597
unemployed	1101	202	1303
unknown	254	34	288
Total	39919	5292	45211

Marginal probability

p_No = 39919/45211
p_yes= 5292/45211

p_adm = 5171/45211
p_bc=9732/45211
p_ent = 1487/45211
p_hm = 1240/45211
p_manag = 9458/45211
p_ret = 2264/45211
p_semp = 1579/4511
p_ser = 4154/45211
p_std = 937/45211
p_tech = 7597/45211
p_unempl = 1303/45211
p_unkno = 288/45211

p_adm
p_bc
p_ent
p_hm
p_manag
p_No
p_ret
p_semp
p_ser
p_std
p_tech
p_unempl
p_unkno

p_yes

0.11437482028709828

0.21525734887527373

0.03289022582999712

0.027426953617482472

0.20919687686624935

0.882948839884099

0.05007630886288735

0.35003325205054314

0.09188029461856628

0.02072504478998474

0.16803432792904383

0.028820419809338436

0.006370131162770122

0.117051160115901

# Probability of Job with respect to resopnse of call 
contigency_table/45211

response_after_call	no	yes	Total
Job
admin.	0.100418	0.013957	0.114375
blue-collar	0.199597	0.015660	0.215257
entrepreneur	0.030148	0.002743	0.032890
housemaid	0.025016	0.002411	0.027427
management	0.180421	0.028776	0.209197
retired	0.038641	0.011435	0.050076
self-employed	0.030789	0.004136	0.034925
services	0.083696	0.008184	0.091880
student	0.014797	0.005950	0.020747
technician	0.149455	0.018580	0.168034
unemployed	0.024352	0.004468	0.028820
unknown	0.005618	0.000752	0.006370
Total	0.882949	0.117051	1.000000

Probability of eduation with respect to resopnse of call

# Making contingency table JOb with respect to response after call
contigency_table = pd.crosstab(bank_df1.Education,bank_df1.response_after_call,margins=True,margins_name='Total')

contigency_table

response_after_call	no	yes	Total
Education
primary	6260	591	6851
secondary	20750	2452	23202
tertiary	11304	1997	13301
unknown	1605	252	1857
Total	39919	5292	45211

# Probability
contigency_table/45211

response_after_call	no	yes	Total
Education
primary	0.138462	0.013072	0.151534
secondary	0.458959	0.054235	0.513194
tertiary	0.250028	0.044171	0.294198
unknown	0.035500	0.005574	0.041074
Total	0.882949	0.117051	1.000000

Let’s check if we have any statistical patterns in the Data frame (using plots or analysis).

plt.figure(figsize=(10,6),dpi=200)

plt.subplot(321)
sns.boxplot(data=bank_df1,x = 'salary')
plt.subplot(322)
sns.boxplot(data=bank_df1,x = 'balance')

plt.subplot(323)
sns.barplot(data = bank_df1, x = 'marital', y = 'salary', ci = None, palette='bright');
plt.subplot(324)
sns.barplot(data = bank_df1, x = 'marital', y = 'balance', ci = None, palette='bright');
plt.subplot(325)
sns.scatterplot(data =bank_df1, x = 'balance', y= 'customer_age');
plt.subplot(326)
sns.scatterplot(data =bank_df1, x = 'salary', y= 'customer_age');


# plt.tight_layout();

observation = In bar plot we can see that average salary of married customers are 57k and saving is more for single and divorced people